An HPLC method is developed in eight steps: define what it must measure, learn the analytes’ properties, choose a first column, set the mobile phase and pH, run a scouting gradient to choose isocratic or gradient elution, size the injection, resolve the critical pair, then test robustness and set system suitability before validation.
AI Method Builder
Draft a starting method for your separation and refine it — save your work to your account.
What are the eight steps of HPLC method development?
The steps run in order because each one fixes a condition the next one depends on. ICH Q14 sets the minimum for any analytical procedure: identify the attributes to be tested, select a suitable technology and instrument, study the performance characteristics and robustness, and document the procedure with its control strategy.1 The eight steps below are a practical route through that minimum for a reversed-phase HPLC method; the enhanced approach in Q14 adds risk assessment, uni- or multivariate experiments and a control strategy with set points or ranges.1
Table 1 lists the decision each step makes, the evidence that settles it and the point at which to move on.
| Step | Decision | Evidence to collect | Move on when |
|---|---|---|---|
| 1. Define the goal | What the method measures and how well | Analytes, matrix, intended use and the performance each result needs | An analytical target profile is written1 |
| 2. Know the analytes | Separation mode and candidate pH | Structure, pKa, log P, UV spectrum, solubility and stability | The charge state of each analyte at the candidate pH values is known |
| 3. Choose a first column | Phase for the scouting run | Analyte properties from step 2 | A C8 or C18 column, or another mode if step 2 rules reversed phase out2 |
| 4. Set the mobile phase | Organic solvent, buffer and pH | Acetonitrile and methanol screens at two or three pH values | pH chosen relative to the analyte pKa values, with a buffer that can hold it3,4 |
| 5. Isocratic or gradient | Elution mode and composition range | A full-range scouting gradient | The peak span is read against the run time2 |
| 6. Size the injection | Injection volume and sample diluent | Injections at the nominal volume, half and double | Peak width and resolution do not change across the test5 |
| 7. Resolve the critical pair | Which lever changes selectivity | Critical-pair resolution across pH, solvent, column and temperature | The required resolution is met with margin |
| 8. Robustness and system suitability | Parameters to control and test criteria | Small deliberate changes to the method parameters | The outcome is documented and reflected in the control strategy1 |
Figure 1 follows Table 1 step by step.
Step 1: What does the method have to measure, and how well?
Write the answer down before choosing a column or a mobile phase. ICH Q14 calls this summary the analytical target profile: a prospective statement of the method’s intended purpose and the performance criteria its measurements must meet. It drives the choice of technology and is the basis for development; Q14 makes formal documentation of the profile optional, but writing one down is good practice.1 For an HPLC method, answer these questions:
- Which analytes must be separated, and which pair is likely to be hardest?
- What resolution, precision and range does each result need?
- Is the method quantitative or qualitative?
- Which matrix will be analyzed, and what else in it can elute?
- Is the method intended for regulated use?
The answers decide every later step. A method for a regulated assay needs documented robustness and system suitability from the start; a screening method for a synthetic chemist may not.
Step 2: Which analyte properties decide the starting conditions?
Collect polarity (log P), pKa, UV absorbance at the intended wavelength, solubility and stability for every analyte. Polarity predicts reversed-phase retention; very polar compounds that are barely retained even at high aqueous content may need another mode, and the decision is set out in when to switch from reversed-phase HPLC to HILIC. Solubility and stability limit the sample diluent and the run conditions.
pKa matters most for ionizable analytes. An ionizable compound is most retained in its neutral form, so acids are run below their pKa and bases above it.3 Knowing each analyte’s charge state at the candidate pH values tells you where small pH changes will move retention and where they will not.
Step 3: Which column should you try first?
For most small-molecule methods, start with a C18 or C8 reversed-phase column: Dolan estimates that around 70–80% of all methods are developed on one.2 Use the same column for the scouting run in step 5, so that its retention data carry forward.
As LabVeda practice, change the column when pH and solvent have not separated the critical pair. A phase’s selectivity reflects the bonded ligand, the silica substrate and post-bonding treatment such as endcapping.4 Table 2 lists the column questions that arise during development and the LabVeda guide that answers each.
| Question | Starting answer | Covered in depth by |
|---|---|---|
| Which column for the first scouting run? | A C18 or C8 reversed-phase column2 | This guide, step 5 |
| Which bonded phase changes selectivity for a critical pair? | Compare phase mechanisms before ordering columns | How stationary-phase chemistry changes selectivity |
| Is the column the right lever, or pH, solvent or temperature? | Change mobile-phase levers before the column | HPLC selectivity optimization |
| Analytes barely retained in reversed phase? | Weigh aqueous-stable phases against HILIC | When to switch from reversed-phase HPLC to HILIC |
| Impurity method needs a second, different separation? | Choose a column with clearly different selectivity | Choosing an orthogonal HPLC column for impurity profiling |
| Monograph names only a USP L-code? | Same substituent, ranked by equivalency, within USP ⟨621⟩ | Choosing an HPLC column for a regulated method |
The interactions behind C18, C8, phenyl, PFP and HILIC phases are compared in how stationary-phase chemistry changes selectivity, and the order in which to try pH, solvent, temperature and column is set out in HPLC selectivity optimization. For impurity methods that need a second separation, see choosing an orthogonal HPLC column for impurity profiling. When a monograph specifies only a USP L-designation, the rules for choosing and replacing the column are in choosing an HPLC column for a regulated method. Figure 2 summarizes the column step.
Step 4: How do you choose the organic solvent, buffer and pH?
Choose the pH first, then a buffer that can hold it. A buffer should have a pKa within 1.0 unit of the target pH; outside that range its capacity falls and more buffer is needed for the same effect. For reproducible retention, set the mobile-phase pH at least 1.5 units above or below the analyte pKa values. Within 1.0 unit of a pKa, pH gives the most control over selectivity, but small pH errors then cause irreproducible retention.3
For LC–MS, use volatile additives such as formic or acetic acid and their ammonium salts, and keep buffer concentrations at or below 10 mM to avoid ion suppression.3 For UV methods, also check that the buffer’s UV cutoff sits below the detection wavelength.
Screen both organic solvents. A practical screen runs acetonitrile and methanol at mobile-phase pH values near 2, 4 and 6, giving six experiments per column, and re-equilibrates the column properly between runs, about 10 column volumes when TFA is used.4 Compare the chromatograms for efficiency, selectivity, peak shape and resolution before choosing one combination to take into step 5.
Step 5: Should the method be isocratic or gradient?
Let a full-range scouting gradient decide. Dolan’s example uses a 150 × 4.6 mm, 5 µm reversed-phase column, a 5–95% B gradient over about 20 min at 2 mL/min, and a starting composition of at least 5% B.2 Read the retention times of the first and last peaks and divide their difference by the run time. If the peaks occupy less than 25% of the run time, an isocratic method should be possible; above 40%, a gradient is needed; between the two, either may work, depending on the sample.2 Dolan presents this as a guide, not a hard rule. Table 3 collects the scouting-run numbers.
| Item | Starting point | Source |
|---|---|---|
| Column | 150 × 4.6 mm, 5 µm, C8 or C18 | Dolan2 |
| Gradient | 5–95% B in about 20 min at 2 mL/min; start at 5% B or higher | Dolan2 |
| Gradient retention factor k* | 2 to 10; start at 5 | Dolan2 |
| Peaks span less than 25% of the run time | An isocratic method should be possible | Dolan2 |
| Peaks span 25–40% | Either mode may work, depending on the sample | Dolan2 |
| Peaks span more than 40% | A gradient is needed | Dolan2 |
| Re-equilibration between screening runs | About 10 column volumes when TFA is used | Taylor4 |
Figure 3 applies the rule to three example runs. With a 20 min run, peaks from 9.0 to 12.5 min span 3.5 min, or 17.5%: isocratic is worth trying. Peaks from 6.0 to 12.4 min span 6.4 min, or 32.0%: either mode may work. Peaks from 3.0 to 17.0 min span 14.0 min, or 70.0%: use a gradient.
If the scouting gradient wastes time at either end, narrow its range and shorten the gradient time in proportion, so that k* stays constant.2 The full treatment of gradient steepness, range and dwell volume is in the guide to HPLC gradient optimization. When a gradient moves to a column of another length, diameter or particle size, the gradient time, flow rate and injection volume must be rescaled as well, as described in scaling a method to new column dimensions.
Step 6: How large an injection, and in what solvent?
Inject in mobile phase where you can. With mobile phase as the diluent, Dolan’s 15% rule limits the injection to 15% of the volume of the first peak of interest, which keeps the increase in peak width, or the loss of resolution, to about 1%.5 A diluent with a higher percentage of organic solvent than the mobile phase is stronger: inside the injected plug the sample moves faster than normal (for a typical small molecule of about 400 Da, about threefold faster for a 10% higher %B and about tenfold for 20%), so the front of the band runs ahead and the peak broadens.5
There is no fixed volume limit for a stronger diluent. Test it: inject the nominal volume, half and double, and compare peak width and resolution; then use a weaker diluent or a smaller volume if the peaks change.5 The mechanism and the symptoms, including the reversed case in HILIC, are covered in injection solvent effects.
Step 7: How do you get the critical pair resolved?
Identify the critical pair, the two peaks with the lowest resolution, and judge every change by what it does to that pair. Assess resolution (Rs), peak shape (tailing factor or symmetry), signal-to-noise and repeatability together; a change that resolves the pair but makes a peak tail has not finished the job.
As LabVeda practice, work on selectivity before efficiency: change pH, organic solvent and temperature first and the column last, and track the critical pair through each change. How the separation factor α, the plate number N and the retention factor k each enter resolution through the Purnell equation is worked through in HPLC selectivity optimization. If a peak tails, look at pH, buffer concentration, the column chemistry and the injection solvent before anything else.
Step 8: How do you confirm robustness and set system suitability?
Test robustness while the method is still being developed. ICH Q14 states that for most procedures robustness is evaluated during development, and that the outcome should be documented and reflected in the analytical procedure control strategy.1 Make small deliberate changes to the parameters most likely to matter, typically column temperature, flow rate, buffer concentration and pH, and record which ones move the critical pair. The design of these studies is covered in the HPLC method robustness testing guide.
The parameters that matter become the system suitability test. Q14 describes system suitability tests as a way to verify that the measurement system performs as needed and to make unacceptable performance easier to detect.1 The parameters and criteria are explained in HPLC system suitability. For a USP monograph method, system suitability must be met, and changes beyond the adjustments listed in USP ⟨621⟩ require revalidation.6 For validated or compendial methods, the applicable procedure and regulatory framework take precedence over the general rules of thumb given here.
What mistakes slow HPLC method development?
Five habits cost the most time:
- Changing several parameters at once, so that no single change can be judged.
- Ignoring analyte ionization and setting the pH close to a pKa without meaning to.3
- Optimizing retention when the problem is selectivity.
- Injecting in a diluent stronger than the mobile phase without testing the volume.5
- Leaving robustness until validation, when a fragile parameter is costly to fix.1
Working through the eight steps in order, and changing one thing at a time, avoids all five.
Frequently asked questions
How long should a scouting gradient be?
Long enough to keep the average gradient retention factor k* between about 2 and 10. For a 150 × 4.6 mm, 5 µm column at 2 mL/min and a 5–95% B gradient, Dolan’s calculation with k* = 5 gives about 18 min, rounded to 20 min. The gradient time scales with the column volume and inversely with the flow rate, so a shorter or narrower column needs a shorter gradient at the same flow and a lower flow needs a longer one. Recalculate rather than copy the time from another method.2
Should I start method development with acetonitrile or methanol?
Screen both. The organic solvent is one of the mobile-phase levers on selectivity, so the two can change the spacing of a critical pair as well as retention. A practical screen runs each solvent at two or three pH values on each candidate column and compares the chromatograms for efficiency, selectivity, peak shape and resolution.4 As LabVeda practice, choose the combination that resolves the critical pair with the most margin, then confirm it is compatible with the detector and the sample diluent.
What buffer concentration should I use for LC–MS?
Keep it at or below 10 mM and use volatile additives such as formic acid, acetic acid, ammonium formate or ammonium acetate. Higher concentrations suppress ionization: in Lupo and Kahler’s comparison, raising ammonium formate from 2 mM to 10 mM reduced overall intensity by about 9% in positive-ion mode and 48% in negative-ion mode. Choose the buffer for its pKa, within 1 unit of the target pH, before you choose its concentration.3
When is an isocratic method better than a gradient?
When the scouting gradient shows the peaks occupying less than about a quarter of the run time, an isocratic method should be possible; when they spread over more than about 40% of the run time, a gradient is needed.2 An isocratic method is usually simpler to transfer, because it does not depend on the instrument’s dwell volume or on re-equilibration between injections, while a gradient elutes widely spread peaks in a reasonable time without losing the early ones.
When should method development move to a different column?
As LabVeda practice, move to a different column when changes to pH, organic solvent and temperature have not resolved the critical pair with enough margin, or when retention or peak shape on the current phase cannot be fixed by the mobile phase. Changing the column is a bigger step than changing the mobile phase, because the new column needs its own screen, so make it after the mobile-phase levers have been explored and with a phase chosen for the interaction the pair needs. A phase’s selectivity reflects its bonded ligand, silica and endcapping.4
Does robustness testing have to wait for validation?
No. ICH Q14 places robustness evaluation in development for most procedures, so that its outcome can shape the control strategy and the system suitability test before validation starts. Testing robustness late means a fragile parameter is discovered when the method is already written, and fixing it then can mean repeating the validation work.1
The takeaway
Method development is fastest when each step settles one decision with evidence: a written target, known analyte properties, a C18 or C8 first column, a pH chosen against the pKa values, a scouting gradient read for its peak span, a tested injection, a resolved critical pair and robustness evaluated before validation. Change one parameter at a time, and let the critical pair decide.
References
- International Council for Harmonisation. ICH Q14 Analytical Procedure Development. Step 4, 1 November 2023. https://database.ich.org/sites/default/files/ICH_Q14_Guideline_2023_1116.pdf (accessed 2026-10-22) — analytical target profile (prospective summary of intended purpose and performance criteria; drives the choice of technology); minimal and enhanced approaches (section 2.1); robustness evaluated during development, documented and reflected in the control strategy (section 5.1); purpose of system suitability tests (glossary).
- Dolan JW. Making the most of a gradient scouting run. LCGC Europe. 2013;26(1). https://www.chromatographyonline.com/view/making-most-gradient-scouting-run (accessed 2026-10-22) — 70–80% of methods developed on C8 or C18; scouting run on 150 × 4.6 mm, 5 µm at 2 mL/min, 5–95% B, start at ≥ 5% B; k* 2–10, start at 5 (≈ 18 → 20 min); peak span < 25% of run time isocratic, > 40% gradient, 25–40% either; trimming the gradient at constant k*.
- Lupo S, Kahler T. New advice on an old topic: buffers in reversed-phase HPLC. LCGC North America. 2017;35(7). https://www.chromatographyonline.com/view/new-advice-old-topic-buffers-reversed-phase-hplc (accessed 2026-10-22) — buffer pKa within 1.0 unit of the target pH; UV cutoff; mobile-phase pH at least 1.5 units from analyte pKa values; most selectivity control, and least reproducibility, within 1.0 unit; neutral forms most retained; volatile additives and ≤ 10 mM for LC–MS; ammonium formate 2 → 10 mM intensity loss.
- Taylor T. The LCGC Blog: practical HPLC method development screening. LCGC. 11 April 2016. https://www.chromatographyonline.com/view/lcgc-blog-practical-hplc-method-development-screening (accessed 2026-10-22) — phase selectivity reflects ligand, silica substrate and post-bonding treatment such as endcapping; screen acetonitrile and methanol at pH ≈ 2, 4 and 6 (six experiments per phase); about 10 column volumes of re-equilibration between runs when TFA is used; compare efficiency, selectivity, peak shape and resolution.
- Dolan JW. How much can I inject? Part II: injecting in solvents other than mobile phase. LCGC North America. 2014;32(11). https://www.chromatographyonline.com/view/how-much-can-i-inject-part-ii-injecting-solvents-other-mobile-phase (accessed 2026-10-22) — stronger diluent = higher %B than the mobile phase; ≈ 3× faster for +10% B (≈ 400 Da molecule), ≈ 10× for +20% B; front of the band pushed ahead; 15% rule for injection in mobile phase (≈ 1% width or resolution change); test at nominal, half and double volume; weaker diluent or smaller volume.
- United States Pharmacopeial Convention. USP General Chapter ⟨621⟩ Chromatography, Stage 4 harmonized text (official 1 December 2022). https://www.usp.org/sites/default/files/usp/document/harmonization/gen-chapter/harmonization-november-2021-m99380.pdf (accessed 2026-10-22) — adjustments of chromatographic conditions; changes other than those listed require revalidation; system suitability must be met.
Further reading
- Snyder LR, Kirkland JJ, Glajch JL. Practical HPLC Method Development. 2nd ed. Wiley.
- Snyder LR, Kirkland JJ, Dolan JW. Introduction to Modern Liquid Chromatography. 3rd ed. Wiley.
- Dong MW. Modern HPLC for Practicing Scientists. Wiley.
Reviewed against cited sources. Development requirements follow ICH Q14 (Step 4, 2023); buffer and pH guidance follows Lupo and Kahler (2017); scouting-gradient guidance follows Dolan (2013); screening guidance follows Taylor (2016); injection guidance follows Dolan (2014); compendial statements follow the Stage 4 harmonized text of USP ⟨621⟩. The Figure 3 runs are hypothetical and the percentages are LabVeda calculations. For validated or compendial methods, the applicable procedure and regulatory framework take precedence over the general rules of thumb given here. Evidence review: October 2026.